When trying to start an agent on my multi repo environment, I get a 500 error from the API, resulting in a “Failed to create agents” message displayed.
The following calls are getting 500 from cursor API:
Looks like there was a brief backend issue this morning that affected Cloud Agents startup. Multi-repo startup is more sensitive to these hiccups than single-repo, which explains the difference in behavior. It’s mostly recovered now, which matches your “broke in the morning, then worked, then broke again”.
Can you try starting the multi-repo agent one more time? If you hit a 500 or “Failed to create agents” again, please send:
the approximate time of the failed start, including your timezone
the Request ID from the failed start in the agent chat window: top-right menu > Copy Request ID
With that, we can look up the exact failure. Let me know how it goes.
Thank you so much for your reply. It indeed worked again earlier this afternoon. I just experienced the bug again in a pretty consistent way for around 10min, and now it works again. Sadly I don’t have a request Id as the agent page was showing this:
The first agent that worked when I was brute testing the agent was this one, so most likely the previous one was 2-3s before bc-f7d9b470-f759-4d7a-a5dd-4a277a0930b5. I would say around 15:12 UTC
Hey Nicolas, thanks, that’s enough. The bcId and the rough time around 15:12 UTC are enough to track down the failed start even without a Request ID.
On the backend, we sometimes see Cloud Agent startup failures, and multi-repo is more sensitive to them than single-repo, which is why single-repo stays stable. That also explains the short 10-minute windows where startup fails with “Failed to create agents” or “Cloud Agent not found”, then recovers on its own.
I’ve logged your details internally so the team can check the server side. I can’t share an exact fix timeline yet. For now, if you hit that window again, retrying after a couple of minutes usually works.
If it starts happening more often or lasts longer than 10 minutes, send a new time including the timezone and the bcId here, and I’ll update the report.
